A Locust Swarm Recognition Algorithm Based on Faster-RCNN and GCN

In recent years, due to the frequent occurrence of locust disasters, locusts, as one of the most harmful pests in the world, can cause large-scale erosion to food crops, so the control of locusts is particularly important. Because of the strong aggregation of locusts, it is difficult to identify a single locust to solve the problem of effective control of locust disasters. In this paper, a locust recognition model based on deep learning network is established. Using the network structure of fast region based convolutional neural network (Faster-RCNN) and graph convolutional network (GCN), a locust recognition model is established. By acquiring locust images in simple and complex environments and training the model, individual locusts and locust groups can be accurately recognized. Compared with common image recognition methods, the recognition efficiency of this algorithm is higher.